Model V3 has leaned toward the home side by roughly two expected goals in its last three simulation runs — a gap driven less by overall form than by a specific matchup: the visitors' full-backs have been targeted for crosses into the box in every away match this season, and the model has learned that the home side's wide forwards convert those situations at an above-average rate.

It isn't a guaranteed outcome, just a probability skew — treat the model's output as one input among several, not a result prediction.